inspect.getframeinfo always uses f_lineno even when
passed a tb. In practice it is not always true that

tb.tb_frame.f_lineno==tb.tb_lineno

so user functions like inspect.getinnerframes (and
hence cgitb) will get wrong information back sometimes.

I fixed this using the attached patch.

This script illustrates via cgitb.
############
def raise_an_error():
    a = 3
    b = 4
    c = 0
    try:
        a = a/c
    except:
        import sys, cgitb, traceback, inspect
        tbt,tbv,tb = sys.exc_info()
        print
'traceback\n',''.join(traceback.format_exception(tbt,tbv,tb))
        print '\n\ncgitb\n',cgitb.text((tbt,tbv,tb),1)

raise_an_error()
############

Logged In: YES 
user_id=6946

Whoops tabs and all that the above script should look like
#####start
def raise_an_error():
    a = 3
    b = 4
    c = 0
    try:
        a = a/c
    except:
        import sys, cgitb, traceback, inspect
        tbt,tbv,tb = sys.exc_info()
        print \
'traceback\n',''.join(traceback.format_exception(tbt,tbv,tb))
        print '\n\ncgitb\n',cgitb.text((tbt,tbv,tb),1)

raise_an_error()
#####finish


The same problem occurs with inspect.trace. this code
#####start
import inspect
try:
    raise ValueError
except:
    print inspect.trace()
#####finish produces
[(<frame object at 0x007BFC60>, 'C:\\tmp\\ttt.py', 5, '?',
['    print inspect.trace()\n'], 0)]

ie the returned trace doesn't reflect the current traceback,
but rather the current frame location. This is also fixed by
the same patch. Still present in 2.4a0. This is a semantic
issue. The current test_inspect.py insists that the above
behaviour is correct so it will also need modifying if the
fix goes in.

Logged In: YES 
user_id=6946

python-2.4a0-954364.patch affects inspect.py,
test_inspect.py fixes this for me in 2.4a0 and 2.3.3, but
the semantics of inspect.trace changed so that
test_inspect.py must alter. If inspect.trace() should always
return something starting at the call of inspect.trace then
this patch is *WRONG*

Logged In: YES 
user_id=11375

In the test program, the exception is raised at line 3, but the .trace() call 
says it was at line 5.

Logged In: YES 
user_id=11375

Alternative version of the patch.

Logged In: YES 
user_id=11375

Applied to both CVS HEAD and the 2.3 branch.  Thanks!